Deep Learning for Signals in MATLAB
Contact us to scheduleCourse Details
This one-day course provides a comprehensive introduction to practical deep learning for signals. Attendees will learn how to create, train, and evaluate various kinds of deep neural networks for signal processing using MATLAB®.
Topics include:
Topics include:
- Importing and labeling signal data
- Using convolutional neural networks for signal classification
- Using recurrent neural networks for signal analysis
- Applying deep learning for anomaly detection
- Improving the performance of a network by modifying training options
- Using apps for interactive workflows
Day 1 of 1
Signal Importing, Labeling, and Management
Objective: Import and organize signal data in MATLAB and preprocess it for analysis, including handling missing values, labeling, and extracting regions of interest.
- Store data using MATLAB data types (e.g., timetable)
- Import data with signal datastores
- Use the Signal Labeler app
- Label region of interest based on time and time-frequency representations
- Automate signal labeling with custom functions
Time-Frequency Transforms and Convolutional Neural Networks
Objective: Use convolutional neural networks and transfer learning to classify observations based on their time-frequency content.
- Visualize deep learning networks
- Create time-frequencey images using the spectogram
- Create training and validation sets
- Augment signals
- Use transfer learning
Custom Networks and Feature Extraction
Objective: Use long short-term memory (LSTM) networks and autoencoders to perform classification and anomaly detection.
- Automatically generate features through wavelet scattering
- Classify signals using LSTMs
- Detect anomalies using autoencoders
- Speed up signal processing functions using GPUs
- Use the Experiment Manager app
Level: Intermediate
Prerequisites:
MATLAB Fundamentals, and some knowledge of signal processing and machine learning concepts. No prior knowledge of deep learning is needed for this course.
Duration: 1 day
